Earth and Planetary Science Letters 444 (2016) 34–44

Contents lists available at ScienceDirect

Earth and Planetary Science Letters

www.elsevier.com/locate/epsl

The impact of glacier retreat from the on local climate: Characterization of mineral dust in the Taylor Dome ice core, East ∗ S.M. Aarons a, , S.M. Aciego a, P. Gabrielli b,c, B. Delmonte d, J.M. Koornneef e, A. Wegner a,f, M.A. Blakowski a a Glaciochemistry and Isotope Geochemistry Lab, University of Michigan, 1100 N. University Ave, Ann Arbor, MI, 48109, USA b Byrd Polar and Climate Research Center, The Ohio State University, 108 Scott Hall, 1090 Carmack Road, Columbus, OH, 43210, USA c School of Earth Sciences, The Ohio State University, 275 Mendenhall Laboratory, 125 South Oval Mall, Columbus, OH, 43210, USA d Disat, University of Milano-Bicocca, Piazza della Scienza 1, Milan, 20126, Italy e Vrije University Amsterdam, de Boelelaan 1085, 1081HV Amsterdam, The Netherlands f Stiftung Alfred-Wegener-Institut für Polar- und Meeresforschung, Am Alten Hafen 26, Bremerhaven, 27568, Germany a r t i c l e i n f o a b s t r a c t

Article history: Recent declines in ice shelf and sea ice extent experienced in polar regions highlight the importance of Received 30 June 2015 evaluating variations in local weather patterns in response to climate change. Airborne mineral particles Received in revised form 6 March 2016 (dust) transported through the atmosphere and deposited on ice sheets and glaciers in Antarctica and Accepted 20 March 2016 Greenland can provide a robust set of tools for resolving the evolution of climatic systems through time. Available online xxxx Here we present the first high time resolution radiogenic isotope (strontium and neodymium) data for Editor: M. Frank Holocene dust in a coastal East Antarctic ice core, accompanied by rare earth element composition, Keywords: dust concentration, and particle size distribution during the last deglaciation. We aim to use these Antarctica combined ice core data to determine dust provenance, with variations indicative of shifts in either dust dust production, sources, and/or transport pathways. We analyzed a series of 17 samples from the Taylor ◦   ◦   ice core Dome (77 47 47 S, 158 43 26 E) ice core, 113–391 m in depth from 1.1–31.4 ka. Radiogenic isotopic strontium and rare earth element compositions of dust during the last glacial period are in good agreement neodymium with previously measured East Antarctic ice core dust records. In contrast, the Holocene dust dataset rare earth elements displays a broad range in isotopic and rare earth element compositions, suggesting a shift from long- range transported dust to a more variable, local input that may be linked to the retreat of the Ross Ice Shelf during the last deglaciation. Observed changes in the dust cycle inferred from a coastal East Antarctic ice core can thus be used to infer an evolving local climate. © 2016 Elsevier B.V. All rights reserved.

1. Introduction implied that an ice sheet is capable of changing the position of the subtropical jet, in turn altering storm trajectories (Hall et al., 1996; Fluctuations in the amount and/or extent of sea ice and ice Kageyama and Valdes, 2000; Laîné et al., 2008; Rivière et al., 2010). shelves alter wind speed and direction, as well as local storm The change in storm trajectories will undoubtedly result in changes trajectories (Vihma, 2014; Francis et al., 2009). Modeling the inter- in precipitation pathways and the ice accumulation rates over ice action between ice sheets and local climate demonstrates that air sheets (Beghin et al., 2014). It is possible that variations in the ex- is cooled locally over an ice sheet, affecting the atmospheric flow tent of glaciation, sea-ice and ice shelves are capable of driving response (Liakka and Nilsson, 2010). The high albedo and altitude significant atmospheric climatic variations on seasonal, decadal, of ice sheets can induce zonal anomalies in surface temperature, millennial and glacial–interglacial cycles. which can modify large-scale atmospheric circulation (Cook and Ice cores from the Antarctic ice sheet provide records of past Held, 1988; Beghin et al., 2014). Furthermore, several studies have climate extending over hundreds of thousands of years (Jouzel et al., 1995; Petit et al., 1999). Chemical and mineralogical character- * Corresponding author. ization of dust particles transported through the atmosphere and E-mail address: [email protected] (S.M. Aarons). deposited on ice sheets and glaciers allow for the reconstruction http://dx.doi.org/10.1016/j.epsl.2016.03.035 0012-821X/© 2016 Elsevier B.V. All rights reserved. S.M. Aarons et al. / Earth and Planetary Science Letters 444 (2016) 34–44 35 of regional and global climate patterns. The dust concentration and composition of long timescale ice cores varies with air temperature as recorded by stable isotopes: previous studies have established that dust concentration is one-to-two orders of magnitude greater during glacial versus interglacial periods. The increased dust depo- sition may be attributed to higher dust availability at source areas and higher wind speeds caused by a stronger equator to pole tem- perature gradient (Hammer et al., 1985; Delmonte et al., 2004a), or stronger wind gusts in dust source regions during periods of steepened meridonal temperature gradients (McGee et al., 2010). Provenance of dust deposited in ice can be characterized using radiogenic isotopes, including strontium (87Sr/86Sr), neodymium (143Nd/144Nd), hafnium (176Hf/177Hf) (Grousset and Biscaye, 2005; Basile et al., 1997; Delmonte et al., 2004a, 2004b, 2008; Lupker et al., 2010), and REE concentration (Wegner et al., 2012). The iso- topic composition of ice core dust compared to Potential Source Areas (PSAs) of windborne material indicate variations in the dust Fig. 1. Map of Taylor Dome and surrounding area, with major ice core drilling provenance (Delmonte et al., 2004a; Wolff et al., 2006), which may sites and the hypothesized Last Glacial Maximum (top dashed arrow) and current be used to resolve past climate changes. The longest timescale ice Holocene (bottom dashed arrow) storm trajectory (figure adapted from Morse et al., core records are from interior East Antarctica (Petit et al., 1999), 1998). yet ice cores from coastal East Antarctic sites (i.e. Talos Dome) are capable of providing undisturbed, detailed records of the last climatic cycle in a region of the East Antarctic ice sheet with dis- Sound area over the period 11.5–6 ka (Anderson et al., 1992; tinctive climate conditions (Delmonte et al., 2010). Licht et al., 1996), aligned with the proposed timeline for southerly Previous work on central and coastal east Antarctic ice cores storm trajectories. identify southern South America (SSA) as the most likely source We utilize the TYD ice core dust record to examine the changes of windblown mineral dust during late Quaternary glacial periods. in local weather and storm trajectories during and following the The size distribution and chemical composition of dust from inter- retreat of the Ross Ice Shelf during the last deglaciation. Dry and glacial periods is more variable and in coastal ice cores may poten- windy conditions throughout the LGP are likely to have established tially originate from local sources, although analytical limitations the dominance of dry-deposited dust on TYD, as opposed to sea- have made correlating interglacial ice core dust to source area salt aerosol. The dust accumulation rates should have decreased challenging (Delmonte et al., 2007, 2010; Gabrielli et al., 2010). throughout the last deglaciation, and the sources and transport In 1992, a ∼554 m deep ice core was retrieved at Taylor Dome pathways of dust may have remained similar to those from the LGP ◦   ◦   (TYD) (M3C1 ice core, 77 47 47 S; 158 43 26 E, 2365 m a.s.l.), a (Hinkley and Matsumoto, 2001)or alternatively could have been local ice-accumulation area for the Taylor Glacier. TYD is located completely restructured due to the major climate shift. The retreat on the eastern margin of the East Antarctic ice sheet (Fig. 1), in of the Ross Ice Shelf, 11.5–6 ka, should have had a significant im- close proximity to the current position of the Ross Ice Shelf and pact upon the dust record, consistent with the dust record at Talos Dome (Delmonte et al., 2010)ashypothetically, storm trajectories seasonal sea ice of the Ross Sea. The TYD core was the third and dominant winds would migrate southward to approach TYD ice core (following Vostok and the first Dome C core) to pro- from the southeast (Fig. 1). vide a stratigraphically intact record of the Holocene through the Sea salt aerosol, is yet another indicator of sea ice and ice shelf last glacial cycle, going back to ∼130 ka (Grootes et al., 1994; extent (Wolff et al., 2003; Lupker et al., 2010); it originates from Steig et al., 2000). sea ice covered with brine, frost flowers and bubble bursting over Katabatic-driven movements of cold interior Antarctica air seawater. Here we measure the isotopic and elemental character- masses approaching from the southwest largely influences the istics of the soluble fraction (<0.2μm) of the TYD ice core for weather at TYD (Morse et al., 1998), whereas warmer, precipita- comparison to the insoluble fraction, which is comprised of dust tion-laden air masses approach TYD from the south (Morse, 1997). between 0.2 and 30 μm in diameter. Particles larger than 30 μm in The latter air masses are linked to cyclones originating near Marie diameter were not measured due to the low dust availability and Byrd Land (Fig. 1), traveling over the Ross Ice Shelf and across the instrumental detection limitations. before deposition at TYD (Harris, 1992). This work presents the first detailed Sr and Nd isotopic dataset During the Last Glacial Maximum (LGM), the accumulation rate of coastal Antarctic ice core dust during the last deglaciation. We at TYD drastically decreased, suggesting a change in atmospheric employ a newly developed mass spectrometry technique utilizing circulation during the late ice age period (Morse et al., 1998). 1013 Ohm resistors which can effectively measure variations in Morse et al. (1998) hypothesized that moisture-bearing storms ar- Nd isotope composition of extremely small samples to the fourth rived at TYD from the north (rather than the south as they do decimal place (Koornneef et al., 2014). The goals of the study presently), a result of changing ice cover in the Ross Embayment. are: (1) to provide the first high-resolution record of dust de- The elevated topography in the Ross Embayment (e.g. from the Ice posited in East Antarctica during the last deglaciation and into the Shelf or Ross ‘Ice Sheet’) and northward displacement of the Ross Holocene, (2) identify and compare the geochemical (radiogenic Low combined to displace the storm tracks northward through the isotope compositions and rare earth element concentrations) and Transantarctic Mountains north of the Royal Society Range (Morse physical characteristics (dust concentration and particle size) of ice et al., 1998)(Fig. 1). Terrestrial and marine geological evidence in- core dust to PSAs from SSA and the Ross Sea sector (e.g. McMurdo dicates that grounded ice advanced far into the Ross Sea during the Dry Valleys), (3) explore the effects of the retreating Ross Ice Shelf Last Glacial Period (LGP), reaching its greatest thickness and extent upon regional storm trajectories using the dust record preserved between 12.8–18.7 ka (Hall et al., 2015). The recession of the ice in the TYD ice core, and (4) establish the likely sources of dust to sheet began about 12.8 ka, and it retreated from the McMurdo TYD during the LGP and the Holocene. 36 S.M. Aarons et al. / Earth and Planetary Science Letters 444 (2016) 34–44

2. Materials and methods

2.1. Ice core processing

A series of 17 ice core samples from the TYD ice core (M3C1) were selected between 113 and 391 m depth (see Table 1 for depths and ages). The samples measured here encompass a time period of ∼1.1 to ∼31.4 ka before 1950 A.D. Synchronization of the CO2 concentration record to EPICA Dome C (EDC) and Dronning Maud Land (EDML) ice cores provided chronological refinement of the TYD ice core to approximately 20 ka (Monnin et al., 2004). Re- cent work has refined the TYD age-scale (Baggenstos, 2015)based on combining the work of Monnin et al. (2004) to 20 ka, and Ahn and Brook (2007) to 60 ka. The updated age scale of Baggenstos (2015) is indistinguishable from the original Brook et al. (2000) age scale for 20–40 ka, the time period encompassing the oldest sample in this study. We use the Monnin et al. (2004) timescale for 1–20 ka and Brook et al. (2000) for 20–60 ka. Each sample is approximately 660 g and 22 cm long and spans between ∼3 and ∼30 yrs. The ice was cut into 3 longitudinal samples for radiogenic isotope, REE, and dust concentration and size distribution analysis, respectively. The REE and dust concen- tration and size distribution samples were cut latitudinally into two subsamples (labeled ‘a’ and ‘b’, see Tables S1–S4) to obtain a higher temporal resolution. Decontamination and melting occurred in class 10 laminar flow hoods at the University of Michigan. To remove the outer layer and potential contamination each ice sam- Fig. 2. Schematic of ice core sample preparation for radiogenic isotope and rare ple was scraped using acid-cleaned PFA chisels. The sample was earth element analysis of insoluble and soluble portion of TYD ice core. ∗ Insoluble then rinsed using ultra-pure distilled ethanol (Acros Organics) and portion of ice core is mineral dust and soluble portion is filtered ice, denoted as ‘F’ subsequently rinsed twice with MilliQ water. and ‘W’ respectively.

2.1.1. REE and dust concentration sample preparation procedures of Aciego et al. (2009) and Aarons et al. (2013) for Sr The traditional “acid leach” REE portion of each sample was and Nd isotopic analysis. For a full description of sample prepara- triple rinsed with MilliQ water using acid pre-cleaned LDPE tion see Fig. 2. calipers and melted in pre-cleaned LDPE Nalgene bottles follow- ing procedures established by Boutron et al. (1990). The meltwater 2.2. PSA sample processing was immediately acidified in 1% HNO3 (ultra-pure) for approxi- mately 1 month prior to analysis. The “full digestion” REE portion consisted of 100 μL taken from each insoluble and soluble sample We use PSA samples from the East Antarctic ice sheet mar- digested and dissolved in 1ml of 6M HCl. The samples were sub- gin for source to sink analyses of TYD ice core dust to PSA dust sequently dried down and acidified in 1% v/v HNO3 for comparison from the Ross Sea Region. The Ross Sea Region PSAs were selected to the traditional “acid leach” REE portion of ice core samples. from the fine fraction of regolith, exposed glacial deposits, and The dust concentration and size distribution samples were sediments below the surficial deflation eolic pavement (1–3 cm) triple rinsed with MilliQ water using acid pre-cleaned LDPE (Blakowski et al., in press). Approximately 10 mg of each sample calipers and stored frozen in triple rinsed PFTE centrifuge tubes were digested in concentrated HF in a Savillex Teflon beaker inside ◦ until just prior to analysis following the procedures described in a steel-jacketed Parr acid digestion bomb at 220 C for 48 h. The Delmonte et al. (2004a). insoluble PSA samples were then dried down and immersed in 6M ◦ HCl at 180 C for 12–24 h, dried down, and dissolved in 1mL of 2.1.2. Radiogenic isotope sample preparation 9M HCl. Sample aliquots of 100 μL were extracted and dried down Ice core sections for radiogenic isotope analysis were buffered prior to being re-acidified in 1% HNO3. The acid leach method for to neutral pH during melting with ultra-pure optima grade ammo- the ice samples is: acidification in 1% HNO3 (ultra-pure) for ap- nia (Fischer Scientific) to prevent the dissolution of mineral parti- proximately 1month prior to analysis consistent with the methods cles (Aciego et al., 2009). Immediately afterwards, melted samples from Uglietti et al. (2014). were filtered through pre-cleaned 0.2 μm and 30 μm filters (parti- cles greater than 30 μm were not measured due to dust availability 2.3. REE concentration analysis and analytical limitations). The filtered water (we operatively de- fine as the “soluble fraction” and denote as ‘W’) was collected in a pre-cleaned Teflon beaker for analysis. The REE concentrations of TYD ice were measured using ICP- Insoluble fractions (denoted as ‘F’) were dissolved directly off SFMS (Element 2, Thermo Scientific) coupled with a micro-flow the 0.2μmfilters using ultra-pure HCl, aqua regia, and HNO3–HF nebulizer and desolvation system (Apex Q) in the clean laboratory acid for “total digestion” portions, whereas the soluble fraction was at the Byrd Polar and Climate Research Center at The Ohio State dried under nitrogen flow and infrared radiation before dissolution University following the procedures in Gabrielli et al. (2006, 2010). − in ultra-pure 9M HCl acid for chemistry (Aciego et al., 2009). REE detection is at the sub-pg g 1 levels (Gabrielli et al., 2006), Finally, aliquots of both soluble and insoluble fractions (denoted with procedural blanks concentrations 61 and 7.5 times smaller as ‘W’ and ‘F’ respectively) were also chemically separated using than sample concentrations for the traditional leaching and full -ion-exchange columns and Eichrom resins following established acid digestion method respectively (see Table S3). S.M. Aarons et al. / Earth and Planetary Science Letters 444 (2016) 34–44 37

Table 1 87 86 Radiogenic isotope compositions of Taylor Dome ice samples. εNd and Sr/ Sr isotopic compositions of ice core samples measured in this study, top depth and age are noted. Ages are based upon synchronization of CO2 record in TYD, EDC and EDML ice cores (Monnin et al., 2004). ‘F’ and ‘W’ are insoluble and soluble portions.

87 86 −6 14i 144 −6 −6 Sample ID Top depth Ice age Sr/ Sr ± 2σ 10 Nd/ Nd ± 2σ 10 εNd ± 2σ 10 (m) (ka) 113F 113.12 1.1 0.712856 (227) 0.512296 (129) −6.7 (2.5) 113W 113.12 1.1 0.710033 (65) 0.512286 (191) −6.9 (3.7) 136F 136.12 1.5 0.708418 (540) – – 136W 136.12 1.5 0.710078 (86) 0.511948 (126) −13.5 (2.5) 176F 176.10 2.2 0.708323 (297) – – 176W 176.10 2.2 – 0.512297 (130) −6.7 (2.5) 208F 208.13 3.1 0.714972 (115) 0.512436 (191) −3.9 (3.7) 208W 208.13 3.1 0.709012 (50) 0.512111 (277) −10.3 (5.4) 266F 264.80 4.8 0.705428 (33) 0.512446 (163) −3.8 (3.2) 266W 264.80 4.8 0.705152 (33) 0.512368 (137) −5.3 (2.7) 281F 279.48 5.3 0.716308 (259) 0.512273 (75) −7.1 (1.5) 281W 279.48 5.3 0.705353 (21) 0.512360 (84) −5.4 (1.6) 296F 294.68 6.0 0.706593 (46) – – 296W 294.68 6.0 0.710128 (33) 0.512242 (89) −7.7 (1.7) 299F 297.50 6.1 0.709332 (178) – – 299W 297.50 6.1 0.710383 (57) 0.512282 (74) −6.9 (1.4) 308F 306.50 6.5 0.710751 (244) 0.512594 (170) −0.9 (3.3) 308W 306.50 6.5 0.707023 (38) 0.512347 (62) −5.7 (1.2) 323F 321.5 7.3 0.709708 (168) 0.512550 (87) −1.7 (1.7) 323W 321.5 7.3 0.709522 (75) 0.512333 (93) −6.0 (1.8) 340F 339.14 9.4 0.713616 (229) 0.512509 (132) −2.5 (2.6) 340W 339.14 9.4 0.710054 (64) 0.512496 (71) −2.8 (1.4) 349F 347.6 10.5– – – 349W 347.6 10.5 0.709885 (38) 0.512227 (86) −8.0 (1.7) 366F 364.60 13.0 0.712416 (98) 0.512398 (194) −4.7 (3.8) 366W 364.60 13.0 0.709389 (72) 0.512588 (165) −1.0 (1.4) 376F 374.70 15.5 0.705225 (28) 0.512544 (47) −1.8 (0.9) 376W 374.70 15.5 0.708497 (74) 0.512595 (54) −0.8 (1.1) 381F 380.3 19.7 0.710116 (24) 0.512505 (8) −2.6 (0.15) 381W 380.3 19.7 0.709255 (39) 0.512427 (53) −4.1 (1.0) 383F 381.5 20.7 0.708974 (9) 0.512459 (62) −3.5 (1.2) 383W 381.5 20.7 0.710059 (5) 0.512583 (25) −1.1 (0.5) 392F 391.12 31.4 0.708714 (35) 0.512591 (71) −0.9 (1.4) 392W 391.12 31.4 0.708950 (77) 0.512605 (47) −0.6 (0.9)

2.4. Dust concentration and size distribution analysis (2015) of 143Nd/144Nd = 0.512637. The 143Nd/144Nd ratio of iso- topic standard CIGO (100 pg) was measured simultaneously at Dust concentration and size distribution measurements were 143Nd/144Nd = 0.511344 ± 49 (2σ SD, n = 5), close to the long- conducted by Coulter® Counter at the University of Milan as in term average value of 143Nd/144Nd = 0.511334 ± 10 (2σ SD, Delmonte et al. (2002) and kept frozen until analysis. For each n = 28) (Koornneef et al., 2014), demonstrating that 1013 Ohm dust concentration and size distribution subsample ∼20 ml was resistors can effectively measure variations in 143Nd/144Nd to the available for Coulter® Counter microparticle concentration and size fourth decimal place for samples as small as 100 pg. Two LGP distribution measurements in the range of 1.006–29.83 μm. samples (381 and 383) were measured on a Thermo Scientific Tri- ton Plus TIMS using standard 1011 Ohm resistors at the University 146 144 = 2.5. Strontium and neodymium isotope analyses of Michigan and normalized to Nd/ Nd 0.7219 using the exponential law and mass 149 was monitored to detect any Sm Chemically separated Sr and Nd were measured using ther- interference. BCR-2 (10 ng) measured concurrently with the sam- ± = mal ionization mass spectrometry (TIMS) equipped with 1011 ples averaged 0.512642 42 (2σ SD, n 4), similar to measured Ohm resistors for 87Sr/86Sr ratios and 1013 Ohm resistors for values (Jweda et al., 2015). The long-term average for JNdi-1 is ± = 143Nd/144Nd ratios. Sr isotopic compositions were measured on 0.512101 12 (2σ SD, n 110). the University of Michigan Thermo Scientific Triton Plus TIMS and normalized to 88Sr/86Sr = 8.375209 to account for mass bias. 3. Results The Sr isotopic standard NBS987 (100 ng) long-term average is 87Sr/86Sr = 0.710245 ± 17 (2σ SD, n = 268). The USGS reference The dust concentration and size distribution, REE, Na and Sr material BCR-2 (10 ng) measured concurrently with samples aver- concentrations of the fifteen ice core sections are summarized in aged 0.705002 ± 37 (2σ SD, n = 3), which agrees with published Tables S1, S3, and S4 respectively. The dust flux calculations and BCR-2 values (0.705013, 0.705000; Aciego et al., 2009 and Jweda the sample information for Ross Sea Region PSAs are presented in et al., 2015 respectively). Tables S2 and S5 respectively. Sr and Nd isotopic composition of The majority of Nd isotopic compositions were measured on the TYD dust and ice samples are summarized in Table 1. a Thermo Scientific Triton Plus TIMS using low-noise 1013 Ohm resistors at the Vrije Universiteit in Amsterdam and normalized 3.1. Dust concentration and size distribution to 146Nd/144Nd = 0.7219 using the exponential law and mass 147 was monitored to detect any Sm interference (Koornneef Dust concentrations are >500 and <15 ppb during glacial and et al., 2014). USGS reference material BCR-2 (1 ng) measured interglacial periods respectively (Fig. S1), which is similar to dust concurrently with the samples averaged 0.512634 ± 32 (2σ SD, concentrations observed in interior Antarctic ice cores (e.g. Del- n = 2), in agreement with measured values in Jweda et al. monte et al., 2002, 2004a, 2007, 2008). In the 1–5 μm, 1–10 μm, 38 S.M. Aarons et al. / Earth and Planetary Science Letters 444 (2016) 34–44

Fig. 3. Rare earth element concentrations of TYD dust in ice. a) Normalized REE concentrations with respect to the mean crustal abundance (Wedepohl, 1995)in the 13 LGP and Holocene acid leached TYD ice samples with REE normalized concentrations of ice from EDC, and EDML also separated by time periods (Gabrielli et al., 2010; Wegner et al., 2012). b) Comparison of average acid leached and fully digested TYD ice. c) Comparison of average fully digested TYD ice to Ross Sea sector potential source area dust (Blakowski et al., in press). and 1–20 μm dust diameter size range, the dust concentrations the TYD and EDC ice cores is substantial, however the overall trend range from ∼9–537 ppb, 14–541 ppb, and 14–544 ppb respec- is similar (Fig. 4). The dust flux in the most recent portion of the tively. TYD record is comparable to WAIS Divide ice core (Koffman et al., The mean mass TYD dust particle diameter calculated using 2014a), however comparison to older sections is not yet available. volume distribution increases with decreasing depositional age (Ta- When compared to the Talos Dome record (Albani et al., 2012), the ble S1). Our measured mean mass TYD Holocene dust particle dust flux is slightly higher in the TYD record during the Holocene, diameter (∼3μm) is larger than the glacial dust particle diame- but the difference in dust flux from the two ice cores between the ter (∼1.8μm). The two oldest glacial samples, from ∼31.4–15.5ka LGP and Holocene is distinctive, most notably in the coarse dust are lognormally distributed with a modal value of ∼2μm, consis- particle portion. tent with the long-range transported dust observed in the Talos Dome ice core (Delmonte et al., 2010; Albani et al., 2012). To ex- 3.2. Rare Earth Element concentrations plore the difference in grain size distributions further, we use the proportion of fine particles (fine particle percentage, or FPP) and 3.2.1. REE concentrations of TYD insoluble fraction coarse particles (coarse particle percentage, or CPP) with respect to REEs serve as a valuable tool for characterizing mineral dust in the total mass as defined in Delmonte et al. (2004b). Samples from Antarctic ice, and we differentiate between light REEs (LREEs: La, the Holocene (∼1.1–10.5ka) on average have a high CPP greater Ce, Pr, Nd), medium REEs (MREEs: Sm, Eu, Gd, Tb, Dy, Ho), and than 2μm(∼85%) than older samples (∼13–31.4ka) with a CPP heavy REEs (HREEs: Er, Tm, Yb, Lu). The REE concentrations of the of ∼69% (Table S1). TYD ice were normalized to the mean upper continental crustal Larger dust particle diameters observed during the Holocene is abundance of each element (Wedepohl, 1995)(Fig. 3). Compara- also observed at Talos Dome, a coastal ice core in East Antarctica ble shapes of REE patterns could indicate uniform dust sources (or (Albani et al., 2012) and the West Antarctic Ice Sheet (WAIS) Divide uniform mixing of sources); whereas departures from a constant deep ice core (Koffman et al., 2014a). The dust input at TYD, Talos pattern could indicate a change in dust source(s). The REE con- Dome and WAIS Divide during the Holocene is dominated by larger centrations of the TYD samples follow the dust concentrations for dust particles whereas the dust concentration during the LGP is glacial/interglacial REE concentrations in other Antarctic ice cores attributed to finer dusts. (Gabrielli et al., 2010; Wegner et al., 2012). Higher input of conti- The dust flux rates at TYD are determined using the variable nental dust during glacial periods results in higher REE concentra- water equivalent accumulation rates (calculated from ice equiva- tions observed in glacial samples compared to interglacial samples. lent) for each time period (Morse et al., 1998) and dust concentra- All of the samples from the Holocene display a unique positive tion (this study) (see Table S2, Fig. 4). On average, the flux of dust Eu anomaly (Fig. 5b), which is not found in Holocene samples in − − particles between 1–20 μm during the LGP was ∼3.9mgm 2 yr 1 ice cores from the interior East Antarctic ice sheet (Gabrielli et al., − − compared to ∼2.3mgm 2 yr 1 during the Holocene. The average 2010; Wegner et al., 2012). flux of dust particles between 1–5 μm and 1–10 μm was ∼3.6 − − and ∼3.7mgm 2 yr 1 during the LGP and decreased to ∼1.1 and 3.2.2. REE concentrations of TYD soluble fraction − − ∼1.5mgm 2 yr 1 during the Holocene (Table S2). The dust flux The REE concentrations of the full-digestion soluble and insolu- record of fine particles (1–5 μm) at TYD follows a similar pattern ble fractions follow similar but not identical patterns (Fig. S4). REE observed at EDC, however the change in TYD dust flux across the concentrations of the LGP ice samples are higher than the majority termination of the LGP is smaller (Fig. 4). The flux of larger dust of the Holocene samples and have a unique positive Eu anomaly. particles (up to 20 μm) is significantly higher during the Holocene Another sample (1.1 ka, sample 113) has a pronounced enrichment compared to the LGP (Fig. 4). The difference in dust flux between in Pr not observed in any other soluble samples (Fig. 6). We would S.M. Aarons et al. / Earth and Planetary Science Letters 444 (2016) 34–44 39

18 Fig. 4. Thirty-two-thousand-year climate history of the Ross Sector of East Antarctica. TYD ice core δ O isotope profile (blue line; Steig et al., 2000), CH4 concentration (red line; Brook et al., 2000), CO2 concentration (grey line; Monnin et al., 2004), Sr and Nd isotopic compositions (brown circles = dust fraction, blue diamonds = ice fraction) and the Sr isotopic composition of modern seawater (solid blue line) (Hodell et al., 1990). TYD dust flux record for three different size fractions (red circles = 1–5 μm, blue circles = 1–10 μm, and black circles = 1–20 μm, this work) with EDC dust flux (black line; Lambert et al., 2012). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.) 40 S.M. Aarons et al. / Earth and Planetary Science Letters 444 (2016) 34–44 expect the REE composition of the soluble fraction of TYD ice to be similar to seawater, however only Antarctic Bottom water REE data exists in current literature (Kawabe et al., 1998). The REE composi- tion of the soluble fraction of TYD ice and Antarctic Bottom water do not match (Fig. 6).

3.2.3. REE concentrations of potential source area dust We include 76 PSA dust samples from the Ross Sea Sector (Baroni et al., 2008; Delmonte et al., 2010)from 17 locations along the north-central Transantarctic Mountains, the ice-free val- leys and nunataks of and nearby McMurdo Sound sector for source-to-sink comparison. The bedrock ranges from late Proterozoic to early Paleozoic granites and gneisses, Devo- nian to Triassic sedimentary rocks and the sills of the Jurassic Ferrar dolerite (Blakowski et al., in press; Stump and Fitzgerald, 1992). The sampled material ranges from regolith, glacial dunes, lacustrine sand, and volcanic material. See Supplementary Table S5 for more detailed information on the sam- ple geologic history, geographic surroundings, climate and collec- tion. The REE concentrations of the PSA dust from the Ross Sea Sector display a uniform crustal pattern (Fig. 3c), with a slight en- richment of heavy REE with respect to light REE.

Fig. 5. Radiogenic isotopic compositions of Taylor Dome ice core dust and regional 3.3. Sr and Nd isotopic compositions Potential Source Areas. Sr and Nd isotopic compositions of Taylor Dome ice core Holocene (black circles) and Glacial dust (white circles) with regional Potential We use the radiogenic isotopic composition of dust entrained Source Areas plotted in various colors (colored crosses are individual data points) in TYD ice to discern the shifts in provenance during the time (Delmonte et al., 2010, 2013; Blakowski et al., in press) and long range Potential Source Area of Southern South America plotted within dashed gray line (Delmonte period studied. The isotope variability in the glacial samples et al., 2004a; Gaiero, 2007). 87 86 is large: Sr/ Sr = 0.7052–0.7101 and εNd =−3.5 to −0.9, whereas Holocene dust displays an even broader range: 87Sr/86Sr = 0.7054–0.7163 and εNd =−7.1 to −0.9. We observe signifi- Holocene. The higher total Holocene dust flux was also observed cant variations in the εNd compositions of the Holocene soluble in the Talos Dome ice core (Albani et al., 2012). This higher input fraction; the insoluble and soluble fractions from a single sample of large dust particles suggests more local input of dust, which can are not coupled and suggest that the variable soluble εNd com- be attributed to an alteration in storm trajectories and dominant positions are not a result of leaching from the insoluble fraction weather pathways (Morse et al., 1998). The ice-free terrains of Vic- (Fig. 4). The majority of soluble samples measure close to the es- torialand and the nearby Transantarctic Mountains during the last tablished 87Sr/86Sr seawater composition of 0.70917 (Hodell et al., glacial period rule out the possibility of increased large dust parti- 1990)with the exception of samples from ∼4.7, 5.3, and 6.5 ka cle concentration due to a reduction of land ice cover and resulting (0.705152, 0.705353, and 0.707023 respectively). Most of the in- exposure of local dust sources (Delmonte et al., 2010). soluble samples possess a higher Sr radiogenic isotope composition The difference in dust flux between the TYD and EDC ice cores than the water-soluble fraction (Fig. 4). is significant (Fig. 4). The overall trends of dust flux in both ice The TYD insoluble portion spans a broad range of Sr and Nd cores are similar: higher (lower) proportion of fine particle flux isotopic compositions (Fig. 5). The majority of the Holocene sam- during LGP (Holocene), though the magnitude of difference is less ples are more radiogenic with respect to Sr (average Holocene = in the TYD core most likely due to the proximity to unglaciated 0.710573 versus average LGP = 0.708257), and plot closer to the terrains. However, due to the discontinuous record of the TYD core, regional Ross Sea Region patterns (colored areas, Fig. 5). However, it is difficult to speculate as to the underlying causes of the dust at least 3 of the Holocene samples plot within the SSA source area. flux differences between the TYD and EDC ice cores. The glacial samples plot further away from the Ross Sea Region The measurement of size variations of dust particles entrained source area pattern (with the exception of a sample from ∼13 ka, in ice cores has proven useful for inferring changes in dust which may be a result of the deglacial transition), also observed in transport pathways and atmospheric circulation (Delmonte et al., dust from a previously measured East Antarctic ice core (Delmonte 2004b). The modal diameter in the TYD glacial samples is ∼2μm, et al., 2004a, 2008). consistent with previous results obtained from another coastal East Antarctic ice core (Talos Dome) (Albani et al., 2012), as well as in- 4. Discussion terior East Antarctic ice cores (Vostok, EDC) (Petit et al., 1999; Delmonte et al., 2004a). The small dust diameters observed in 4.1. Dust concentration, flux and size distribution: evidence of a ice cores across East Antarctica during the LGP are indicative of changing regional climate long-range sourced dust that was most likely deposited congru- ently throughout East Antarctica (Petit et al., 1999; Delmonte et Long-term variations in dust concentration and flux have been al., 2004a; Albani et al., 2012). noted in records from the EDC ice core from interior East Antarc- The modal diameter of Holocene dust in the TYD ice core is tica (Lambert et al., 2008). The EDC ice core records a drastic drop larger at ∼3μm, and the samples are not log-normally distribu- (an order of magnitude) in dust flux (Fig. 4) and dust concen- ted—most likely due to poor size-sorting during transport and tration (Fig. S2) following the termination of the LGP (Delmonte deposition. In contrast, ice cores from interior East Antarctica et al., 2004a). Similarly, the flux, concentration and size distribu- (Dome B, Komsomolskaia, EDC) during the time period of 12 ka tion of TYD ice core dust changes dramatically from the LGP to have smaller average dust particle diameters ranging from 1.8, 1.9, the Holocene, with larger dust particles more prevalent during the and 2.2 μm respectively (Delmonte et al., 2004b). Talos Dome, a S.M. Aarons et al. / Earth and Planetary Science Letters 444 (2016) 34–44 41 coastal ice core, is characterized by poor size sorting during the order of magnitude (as described above in methods), and com- Holocene and the presence of dust particles in the size range of parison of both techniques yielded varying REE concentration 5–10 μm is strong evidence of local input (Albani et al., 2012). patterns (Fig. 3), which also occurred in earlier work measuring Analysis of the TYD dust size distribution in three size fractions trace element concentration (Koffman et al., 2014b). The full di- (1–5, 1–10, and 1–20 μm respectively) highlights the shift in av- gestion Holocene TYD samples all display a unique Eu enrichment erage dust particle diameter between the LGP and the Holocene anomaly, with the exception of the sample from the time period (Fig. S2). of ∼6.1 ka (sample 299). The fully digested LGP TYD samples do not have this positive Eu anomaly, supporting the hypothesis that 4.2. Rare Earth Element signatures of Taylor Dome ice dust deposited on TYD during the dust origin during the LGP is different than during the Holocene. Prior work comparing the acid leaching method to the full acid digestion method determined that The REE patterns normalized to upper continental crustal abun- the traditional acid leaching method overestimates the LREE con- dance are consistent with input of dust from continental sources centrations with respect to the MREE and HREE concentrations (Fig. 5a). The normalization is applied to the samples and results (Gabrielli et al., 2010). In contrast, the leached TYD samples pos- in a characteristic REE profile with a clear positive Eu anomaly for sess lower normalized La concentrations compared to the fully Holocene samples (Fig. 3). Any deviations from this pattern would digested samples (Fig. 3b). The results presented here demonstrate be indicative of a change in source area, which is best determined that the sample methodology of traditional acid leaching versus by comparing these results to the REE profiles of soil and rock from full acid digestion produces distinctly different REE concentrations. hypothesized source areas (see Fig. 3c). The REE patterns (both the acid leached and fully digested 4.2.2. REE comparison of Taylor Dome dust to PSAs samples) for the TYD ice core display a similar, temporally un- Reconstruction of dust sources and pathways through time is changing profile throughout the LGP. There is a distinct change possible through the comparison of REE patterns in ice cores (positive Eu anomaly) for samples younger than ∼15.5 ka (sample to those measured in PSAs (Wegner et al., 2012). We compare 376), although a Holocene sample (∼6.1 ka, sample 299) displays the fully digested acid leach data of TYD dust to Ross Sea sec- a pattern similar to LGP samples (Fig. 3). The positive Eu anomaly tor PSA dust samples as both sets of data were processed the observed in the TYD Holocene samples (Fig. 3a, b) is also observed same. The REE concentrations of TYD LGP samples are distinctly in young volcanic PSA dust from Patagonia (Gaiero, 2004), however different from the REE patterns of the Ross Sea sector PSA sam- it is unlikely that the TYD Holocene dust is originating from Patag- ples (Fig. 3c). The Holocene TYD ice has a significant positive Eu onia as the isotopic compositions are not in agreement (Fig. S7). anomaly not observed in the Ross Sea sector PSA samples; how- The comparison of the results of the two REE methods is described ever, both sets of samples have a similar HREE pattern with the below. exception of the sample from ∼6.1 ka (sample 299) (Fig. 3c). Based on the REE pattern during the Holocene, we hypothesize that the 4.2.1. Comparison of traditional acid leaching to full acid digestion source of dust to TYD during this time period is a consistent mix The results of different sample methodology for REE and trace of dust from PSAs in the Ross Sea Sector and a small but con- element analysis is an important topic still under consideration stant input of long-range transported dust. The REE pattern of (Rhodes et al., 2011; Koffman et al., 2014b). Previous work high- dust deposited in Antarctic ice throughout the last deglaciation lighted the importance of sample methodology upon the mea- suggests an input of a mixed long-range transported dust or in- sured trace element and REE concentrations; different laborato- put from a baseline source such as SSA (Gabrielli et al., 2010; ries may use varying methods for sample preparation and subse- Wegner et al., 2012). The Holocene REE patterns of EDC and EDML quent analysis which may result in incorrect or misleading mea- ice differs significantly (high enrichment in Gd) from those of the sured concentrations due to incongruent elemental dissolution Ross Sea sector PSAs and the Holocene TYD samples, implying that of trace and REEs (Rhodes et al., 2011; Koffman et al., 2014b; the source of dust to East Antarctica during the Holocene changes Uglietti et al., 2014). Previous work using three separate prepa- based upon the proximity to the coast (Delmonte et al., 2013; rations (acid leached, pre-filtered then acid leached, and full di- Bertler et al., 2005)(Fig. 3c). The Holocene TYD REE pattern re- gestion) of Antarctic ice-core samples to determine REE concentra- mains relatively uniform and constant, and may be a reflection tions demonstrated that glacial–interglacial variations of the REE of dust input from a local unrecognized (unmeasured) source. patterns emerge and are significant if the method adopted is con- The current Holocene storm trajectory approaching TYD from the sistently used for all the samples analyzed (Gabrielli et al., 2010). southeast crosses the Royal Society Range, which has not been In our study, the REE patterns of the acid leached and fully di- sampled for chemical characterization of the PSA and therefore gested TYD samples vary from one another: the LREE patterns in cannot be ruled out as a potential contributor of dust to TYD the higher concentrated samples (LGP) are distinctively different, (Fig. 1). whereas the lower concentrated samples (Holocene) all display the We evaluate the possibility of dust contribution from both the positive Eu anomaly (Fig. 3b). Ross Sea sector and SSA PSAs with a simple two-component mix- When using the trace element concentrations to discern ing model of REE concentrations in increments of 10% (i.e. 90% SSA changes in atmospheric trace elements due to anthropogenic pol- dust input and 10% Ross Sea sector dust input) (Fig. S5). The HREE lution and/or land use changes, the sample preparation method enrichment of Ross Sea sector and SSA PSAs is not observed in the becomes more important for comparing ice core records (Koffman average measured Holocene REE concentrations or the modeled et al., 2014b; Uglietti et al., 2014). The results of these studies mixtures of SSA and Ross Sea sector REE concentrations (Fig. S5). emphasize the importance of allowing melted acidified ice core The discrepancy between the observed average TYD Holocene REE samples to sit for at least one month prior to analysis, and that concentrations and the PSAs from the Ross Sea sector and SSA incomplete dissolution of minerals and/or incongruent leaching may be attributed to dust input from an uncharacterized source is an issue that can be resolved by full acid digestion using HF. or fractionation of REEs during long-range transport. Changes in To discern the effects of acid leaching versus full acid digestions the mineral composition may occur during eolian transport due to on our TYD ice core samples we used both techniques (Fig. 3b). gravitational sorting: heavy minerals settle out of the atmosphere Our results demonstrate that full acid digestion results in higher closer to the source, leaving behind suspended particles depleted REE concentrations than the acid leaching method by nearly one in HREE, Zr and Hf (Gaiero, 2007; Aarons et al., 2013). 42 S.M. Aarons et al. / Earth and Planetary Science Letters 444 (2016) 34–44

trations of TYD ice are another indicator of changes in sea salt contribution and we use the Na/Sr ratio to serve as a proxy (Ta- ble S4). The average insoluble and soluble TYD Na/Sr ratios are 191 and 1034 respectively (Fig. S3). The average upper continental crust and seawater Na/Sr ratio is 81 (Wedepohl, 1995) and 1358 (Riley and Tongudai, 1967) respectively, which supports the con- clusion that the higher soluble Na/Sr ratio in the TYD samples is attributable to sea salt acting as the primary constituent (Fig. S3). It is likely that sea salt originating from nearby seasonal sea ice in the Ross Sea is a large contributor to the soluble Sr in the TYD ice, and we therefore interpret our radiogenic isotope data separately as the soluble and insoluble portions.

4.3.1. Radiogenic isotopic composition of soluble portion The water-soluble portions of the TYD ice core plot close to the modern 87Sr/86Sr composition of seawater, and the highest variability occurs between ∼8–4 ka (Fig. 4). It is unclear why three water-soluble samples are not in agreement with the es- tablished 87Sr/86Sr composition of seawater (0.70917, Hodell et al., 1990). However, the variable εNd compositions of Holocene solu- ble samples could be attributed to oceanic processes, the Southern Ocean εNd composition close to the Ross Sea ranges from −11.3to −8.4(Albarède et al., 1997). Aerosols originating from seasonal sea ice with frost flowers during warmer, interglacial conditions may cause the transition from uniform to variable εNd values (Fig. 4).

4.3.2. Radiogenic isotopic composition of insoluble portion The Sr and Nd isotopic compositions of TYD dust indicate a shift provenance from a relatively constant long-range input to a Fig. 6. Rare earth element concentrations of Taylor Dome ice soluble portion (full more variable proximal source following the deglaciation, as the digestion) in comparison to REE concentrations in seawater (Kawabe et al., 1998). Holocene isotopic compositions trend more towards the PSAs from Hollow brown and blue circles are Holocene and Last Glacial Period fully digested samples. Solid dark blue, black and light blue lines are average Pacific seawater, the Ross Sea sector. In general, the Holocene samples span a larger 87 86 Antarctic bottom water, and unique Pacific seawater samples respectively (Kawabe range of Sr/ Sr and εNd isotope values, and appear to be com- et al., 1998). (For interpretation of the references to color in this figure legend, the prised of a combination of Ross Sea sector PSAs. The hypothesized reader is referred to the web version of this article.) Holocene storm trajectory crosses the Royal Society Range (Fig. 1), 87 86 and it has not been sampled or analyzed for Sr/ Sr and εNd iso- 4.2.3. REE concentration of TYD soluble portion topic compositions. The unsampled and unmeasured Royal Society As a first hypothesis, the higher REE concentrations observed Range along with additional unidentified proximal sources cannot during the LGP compared to the Holocene may be attributed to the be ruled out as the predominant source of Holocene dust. The dust increased formation, transport and deposition of sea-salt aerosol from the LGP however, appears to be less radiogenic with respect originating from sea ice, as is the case for other Antarctic ice core to Nd, and lies closer to the SSA source area (Basile et al., 1997)as records (Petit et al., 1999; Wolff et al., 2006). We would expect the concluded by previous studies (Delmonte et al., 2004a, 2010). REE pattern of the soluble LGP samples to most closely resemble the REE pattern of Antarctic seawater, represented by a solid black 4.3.3. Synthesis of physical and chemical characteristics of mineral dust line in Fig. 6. This is not the case. So it is possible that aerosol min- The dust size distribution in TYD ice transitions from fine to eral dust and unbufferable minerals smaller than 0.2μmmay be coarse during the last deglaciation, signifying a transition from dis- present in the soluble portion of the TYD ice core (as the melted tal to local dust sources. The REE concentrations clearly distinguish ice core was filtered to a size of 0.2 μm) along with oceanic derived different dust sources during the LGP and Holocene, and the pres- aerosols (dissolved sea salt), and leached elements from the dust ence of a distinct positive Eu anomaly in almost all of the Holocene fraction during transport and ice core sample processing (Lupker samples suggest that the interglacial samples are originating from et al., 2010). To explore the relative contribution of sea salt ver- a similar source area with congruent bedrock geology. Finally, the sus mineral dust to the soluble REE patterns observed in the TYD radiogenic isotope measurements also imply that the dust sources ice core record, a two-component mixing model in increments of to TYD during the Holocene are similar to local sources (Fig. 5), 10% (i.e. 90% mineral dust and 10% sea salt contribution), is uti- and are most likely a combination between isotopically character- lized (Fig. S6). This two-component model does not produce the ized and uncharacterized areas in the Ross Sea Sector. During the measured REE concentration pattern, which suggests that other LGP we observe isotopic compositions similar to dust from SSA, processes are occurring during either transport or ice formation which supports a shift from long-range transported dust to a local that may fractionate REE patterns. Further, this modeling confirms source. The change in dust sources may be attributed to a change that leaching from mineral dust is unlikely after deposition. of dominant storm trajectories, and is consistent with the hypothe- sis that the moisture-bearing storms approaching TYD have shifted 4.3. Isotopic signature of Taylor Dome and East Antarctic dust from the north to the southeast following the last deglaciation.

Previous work utilizing Sr isotopes as a tracer of aerosol mineral 4.3.4. Retreat of Ross ice shelf reflected in dust record dust highlighted the importance of separating soluble and insol- Glaciological studies (Morse et al., 1998; Hinkley and Mat- uble portions of samples to accurately interpret dust provenance sumoto, 2001) suggest that the Ross Ice Shelf retreat discernable (Aarons et al., 2013). The measured trace element (TE) concen- in paleo-records caused a shift in the local storm trajectories and S.M. Aarons et al. / Earth and Planetary Science Letters 444 (2016) 34–44 43

− the composition of aerosols delivered to the region. Geochemical assistance in the development of 10 13 Ohm resistors for small records of spelothems in northern Europe provide evidence for samples. This research is the Byrd Polar and Climate Research cen- movement in predominant storm trajectories as a result of shifting ter publication 1544. We thank the logistics and drilling team at Late-Pleistocene Northern Hemisphere ice shelf extents (Luetscher Taylor Dome, C. Uglietti for laboratory assistance, and two anony- et al., 2015), thereby confirming that geochemical evaluation of mous reviewers for their helpful comments. paleoclimate proxies may reveal small-scale shifts in climate (e.g. storm tracks). Major and rare earth element concentrations of TYD Appendix A. Supplementary material ice indicate that total impurities were at their highest and dust deposition was greater than sea salt during the LGP (Hinkley and Supplementary material related to this article can be found on- Matsumoto, 2001), perhaps a result of increased dust availability in line at http://dx.doi.org/10.1016/j.epsl.2016.03.035. South America (Basile et al., 1997; Delmonte et al., 2004a, 2010; Gabrielli et al., 2010; Wegner et al., 2012). The sea salt concen- References tration increased in the dust–salt mixture at the termination of Aarons, S.M., Aciego, S.M., Gleason, J.D., 2013. Variable Hf–Sr–Nd radiogenic isotopic the LGP (Hinkley and Matsumoto, 2001), suggesting an increase compositions in a Saharan dust storm over the Atlantic: implications for dust in new sea ice (covered in brine and frost flowers) extent but flux to oceans, ice sheets and the terrestrial biosphere. Chem. Geol. 349–350, unchanged wind strength and patterns as the dust composition 18–26. remained uniform. A reversal in the dust–sea salt concentrations Aciego, S.M., Bourdon, B., Lupker, M., Rickli, J., 2009. A new procedure for separating and measuring radiogenic isotopes (U, Th, Pa, Ra, Sr, Nd and Hf) in ice cores. indicates that the sea ice extent briefly returned to LGP conditions Chem. Geol. 266. between 10–11 ka, before a final retreat and subsequent change to Ahn, J., Brook, E.J., 2007. Atmospheric CO2 and climate from 65 to 30 ka B.P. Geo- sea-salt dominance of aerosol composition. Synthesizing the work phys. Res. Lett. 34 (L10), 703. http://dx.doi.org/10.1029/2007GL029551. of Morse et al. (1998) and Hinkley and Matsumoto (2001) confirms Albani, S., Delmonte, B., Maggi, V., Baroni, C., Petit, J.R., Stenni, B., Mazzola, C., Frez- zotti, M., 2012. Interpreting last glacial to Holocene dust changes at Talos Dome that the Ross Ice Shelf retreat had a profound impact on the local (East Antarctica): implications for atmospheric variations from regional to hemi- climate – changing both storm trajectories and the composition of spheric scales. Clim. Past 8, 741–750. aerosols deposited on the local surroundings, which is supported Albarède, F., Goldstein, S.L., Dautel, D., 1997. The neodymium isotopic composition by the data presented here. of manganese nodules from the Southern and Indian oceans, the global oceanic The broad range of radiogenic isotopic compositions of dust neodymium budget, and their bearing on deep ocean circulation. Geochim. Cos- mochim. Acta 61 (6), 1277–1291. from Holocene ice demonstrates the variable sources of dust to Anderson, J.B., Shipp, S.S., Bartek, L.R., Reid, D.E., 1992. Evidence for a grounded ice TYD during this time period. It is possible that the sources of sheet on the Ross Sea continental shelf during the late Pleistocene and prelimi- dust to TYD during the LGP originated from more than one dis- nary paleodrainage reconstruction. In: Elliot, D. (Ed.), Contributions to Antarctic tal source, however it appears that during the Holocene the dust Research III. In: Antarctic Research Series, vol. 57. American Geophysical Union, Washington, D.C. provenance and dust transport pathways are more variable. Baggenstos, D., 2015. Taylor Glacier as an archive of ancient ice for large-volume samples: chronology, gases, dust, and climate. Ph.D. thesis. University of Califor- 5. Conclusions nia, San Diego. Baroni, C., Fasano, F., Giorgetti, G., Salvatore, M.C., Ribecai, C., 2008. The Ricker Hills Tillite provides evidence of oligocene warm-based glaciation in Victoria Land, We present geochemical and physical measurements of dust in Antarctica. Glob. Planet. Change 60, 457–470. the TYD ice core record spanning the LGP, the deglaciation and Basile, I., Grousset, F.E., Revel, M., Petit, J.R., Biscaye, P.E., Barkov, N.I., 1997. Patag- the Holocene. In addition to providing the first high time resolu- onian origin of glacial dust deposited in East Antarctica (Vostok and Dome C) tion radiogenic isotope compositions of dust in an East Antarctic during glacial stages 2, 4 and 6. Earth Planet. Sci. Lett. 146, 573–589. Beghin, P., Charbit, S., Dumas, C., Kageyama, M., Roche, D.M., Ritz, C., 2014. Interde- ice core, we also utilize and compare two methods of REE sample pendence of the growth of the Northern Hemisphere ice sheets during the last processing (acid leach and full digest). The results presented here glaciation: the role of atmospheric circulation. Clim. Past 10, 345–358. demonstrate that the two sample methodologies produce distinctly Bertler, N., et al., 2005. Snow chemistry across Antarctica. Ann. Glaciol. 41 (1), different REE concentrations, which has implications for studies 167–179. Blakowski, M.A., Aciego, S.M., Delmonte, B., Baroni, C., Salvatore, M.C., Sims, comparing and contrasting REE concentrations in ice core records. K.W.W., in press. Sr–Nd–Hf isotope characterization of dust source areas in The geographic location of TYD relative to the Ross Ice Shelf pro- Victoria Land and the McMurdo Sound sector of Antarctica. Quat. Sci. Rev. vides an ideal study site to explore local shifts in storm trajectories http://dx.doi.org/10.1016/j.quascirev.2016.03.023. related to the rapid retreat of the Ross Ice Shelf during the last Boutron, C.F., Patterson, C.C., Barkov, N.I., 1990. The occurrence of zinc in Antarctic ancient ice and recent snow. Earth Planet. Sci. Lett. 101, 248–259. deglaciation. Our data supports an alteration in local circulation Brook, E.J., Harder, S., Severinghaus, J., Steig, E.J., Sucher, C.M., 2000. On the origin following the termination of the LGP and the subsequent transition and timing of rapid changed in atmospheric methane during the last glacial into the Holocene as evidenced by the shifts in dust particle diam- period. Glob. Biogeochem. Cycles 14, 559–572. eters, REE concentrations, and radiogenic isotope compositions. The Cook, K.H., Held, I.M., 1988. Stationary waves of the ice age climate. J. Climate 1, systematic shift in radiogenic isotope composition combined with 807–819. Delmonte, B., Andersson, P.S., Hansson, M., Schoberg, H., Petit, J.R., Basile-Doelsch, an increase in dust particle diameter during the Holocene suggests I., Maggi, V., 2008. Aeolian dust in East Antarctica (EPICA-Dome C and Vostok): that the dust transported to Taylor Dome is originating from a provenance during glacial ages over the last 800 kyr. Geophys. Res. Lett. 35, local source. These observations are consistent with previous stud- L07703. http://dx.doi.org/10.1029/2008GL033382. ies of a coastal East Antarctic ice core (Delmonte et al., 2010; Delmonte, B., Baroni, C., Andersson, P.S., Narcisi, B., Salvatore, M.C., Petit, J.R., Scarchilli, C., Frezzotti, M., Albani, S., Maggi, V., 2013. Modern and Holocene Albani et al., 2012). We attribute this modification in dominant aeolian dust variability from Talos Dome (Northern Victoria Land) to the inte- weather pathways to the retreat of the Ross Ice Shelf, as first pro- rior of the Antarctic ice sheet. Quat. Sci. Rev. 64, 76–89. posed by Morse et al. (1998). Delmonte, B., Baroni, C., Andersson, P.S., Schoberg, H., Hansson, M., Aciego, S., Petit, J.R., Albani, S., Mazzola, C., Maggi, V., Frezzotti, M., 2010. Aeolian dust in the Ta- los Dome ice core (East Antarctica, Pacific/Ross Sea sector): Victoria Land versus Acknowledgements remote sources over the last two climate cycles. J. Quat. Sci. 25, 1327–1337. Delmonte, B., Basile-Doelsch, I., Petit, J.R., Maggi, V., Revel-Rolland, M., Michard, A., This work was funded by grants from the Rackham Grad- Jagoutz, E., Grousset, F., 2004a. Comparing the Epica and Vostok dust records uate School and the Turner Award from the Department of during the last 220,000 years: stratigraphical correlation and provenance in glacial periods. Earth-Sci. Rev. 66, 63–87. Earth and Environmental Sciences at the University of Michigan Delmonte, B., Petit, J.R., Andersen, K.K., Basile-Doelsch, I., Maggi, V., Lipenkov, V.Ya., to S.M. Aarons and from NSF OPP Antarctic Glaciology Award 2004b. Dust size evidence for opposite regional atmospheric circulation changes 1246702 to S.M. Aciego and P. Gabrielli. We thank C. Bouman for over east Antarctica during the last climatic transition. Clim. Dyn. 23, 427–438. 44 S.M. Aarons et al. / Earth and Planetary Science Letters 444 (2016) 34–44

Delmonte, B., Petit, J.R., Basile-Doelsch, I., Jagoutz, E., Maggi, V., 2007. Late quater- Lambert, F., Delmonte, B., Petit, J.R., Bigler, M., Kaufmann, P.R., Hutterli, M.A., Stocker, nary interglacials in East Antarctica from ice-core dust records. In: Clim. Past. T.F., Ruth, U., Steffensen, J.P., Maggi, V., 2008. Dust-climate couplings over the Interglacials, vol. 7, pp. 53–73. past 800,000 years from the EPICA Dome C ice core. Nature 452, 616–619. Delmonte, B., Petit, J.R., Maggi, V., 2002. LGM-Holocene changes and Holocene Liakka, J., Nilsson, J., 2010. The impact of topographically forced stationary waves millennial-scale oscillations of dust particles in the EPICA Dome C ice core, East on local ice-sheet climate. J. Glaciol. 56 (197), 534–544. Antarctica. Ann. Glaciol. 35, 306–312. Licht, K.J., Jennings, A.E., Andrews, J.T., Williams, K.M., 1996. Chronology of late Francis, J.A., Chan, W., Leathers, D.J., Miller, J.R., Veron, D.E., 2009. Winter northern Wisconsin ice retreat from the western Ross Sea, Antarctica. Geology 24 (3), hemisphere weather patterns remember summer Arctic sea-ice extent. Geophys. 223–226. Res. Lett. 36. http://dx.doi.org/10.1029/2009GL037274. Luetscher, M., Boch, R., Sodemann, H., Spotl, C., Cheng, H., Edwards, R.L., Frisia, S., Gabrielli, P., Barbante, C., Turetta, C., Marteel, A., Boutron, C., Cozzi, G., Cairns, W., Hof, F., Muller, W., 2015. North Atlantic storm track changes during the Last Ferrari, C., Cescon, P., 2006. Direct determination of rare earth elements at the Glacial Maximum recorded by Alpine spelothems. Nat. Commun. 6. http://dx. subpicogram per gram level in antarctic ice by ICP-SFMS using a desolvation doi.org/10.1038/ncomms7344. system. Anal. Chem. 78, 1883–1889. Lupker, M., Aciego, S.M., Bourdon, B., Schwander, J., Stocker, T.F., 2010. Isotopic trac- Gabrielli, P., Wegner, A., Petit, J.R., Delmonte, B., De Deckker, P., Gaspari, V., Fis- ing (Sr, Nd, U and Hf) of continental and marine aerosols in an 18th century cher, H., Ruth, U., Kriews, M., Boutron, C., Cescon, P., Barbante, C., 2010. A major section of the Dye-3 ice core (Greenland). Earth Planet. Sci. Lett. 295, 277–286. glacial interglacial change in aeolian dust composition as inferred from rare McGee, D., Broecker, W.S., Winckler, G., 2010. Gustiness: the driver of glacial dusti- earth elements in Antarctic ice. Quat. Sci. Rev. 29, 265–273. ness? Quat. Sci. Rev. 29, 2340–2350. Gaiero, D.M., 2004. The signature of river-and wind-borne materials exported from Monnin, E., Steig, E.J., Siegenthaler, U., Kawamura, K., Schwander, J., Stauffer, B., Patagonia to the southern latitudes: a view from REEs and implications for pa- Stocker, T.F., Morse, D.L., Barnola, J.-M., Bellier, B., Raynaud, D., Fischer, H., leoclimatic interpretations. Earth Planet. Sci. Lett. 219, 357–376. 2004. Evidence for substantial accumulation rate variability in Antarctica dur- Gaiero, D.M., 2007. Dust provenance in Antarctic ice during glacial periods: from ing the Holocene, through synchronization of CO2 in the Taylor Dome, Dome where in southern South America? Geophys. Res. Lett. 34, L17707. C and DML ice cores. Earth Planet. Sci. Lett. 224, 45–54. http://dx.doi.org/ Grootes, P.M., Steig, E.J., Stuiver, M., Waddington, E.D., Morse, D.L., 1994. A new ice 10.1016/j.epsl.2004.05.007. core record from Taylor Dome, Antarctica. Eos Trans. AGU 75, 225. Morse, D.L., 1997. Glacier geophysics at Taylor Dome, Antarctica. Ph.D. thesis. Uni- Grousset, F.E., Biscaye, P.E., 2005. Tracing dust sources and transport patterns using versity of Washington. Sr, Nd and Pb isotopes. Chem. Geol. 222, 147–167. Morse, D.L., Waddington, E.D., Steig, E.J., 1998. Ice age storm trajectories inferred Hall, B.L., Denton, G.H., Heath, S.L., Jackson, M.S., Koffman, T.N.B., 2015. Accumula- from radar stratigraphy at Taylor Dome, Antarctica. Geophys. Res. Lett. 25, tion and marine forcing of ice dynamics in the western Ross Sea during the last 3383–3386. deglaciation. Nat. Geosci. 8, 625–628. Petit, J.R., Jouzel, J., Raynaud, D., Barkov, N.I., Barnola, J.M., Basile, I., Bender, M., Hall, N.M.J., Valdes, P.J., Dong, B., 1996. The Maintenance of the Last Great Ice Chappellaz, J., Davis, M., Delaygue, G., Delmotte, M., Kotlyakov, V.M., Legrand, Sheets: a UGAMP GCM Study. J. Climate 9, 1004–1019. M., Lipenkov, V.Y., Lorius, C., Pepin, L., Ritz, C., Saltzman, E., Stievenard, M., 1999. Hammer, C., Clausen, H., Dansgaard, W., Neftel, A., Kristinsdottir, P., Johnson, E., Climate and atmospheric history of the past 420,000 years from the Vostok ice 1985. Continuous impurity analysis along the Dye-3 deep core. In: Geophys. core, Antarctica. Nature 399, 429–436. Monog. Series, vol. 33, pp. 90–94. Rhodes, R.H., Baker, J.A., Millet, M.-A., Bertler, N.A.N., 2011. Experimental investiga- Harris, J.M., 1992. An analysis of 5 day midtropospheric flow patterns for the South tion of the effects of mineral dust on the reproducibility and accuracy of ice Pole. Tellus B20, 115–120. core trace element analyses. Chem. Geol. 286, 207–221. Hinkley, T.K., Matsumoto, A., 2001. Atmospheric regime of dust and salt through Riley, J.P., Tongudai, M., 1967. The major cation/chlorinity ratios in sea water. Chem. 75,000 years of Taylor Dome ice core: refinement by measurement of major, Geol. 2, 263–269. minor, and trace metal suites. J. Geophys. Res. 106, 18,487–18,493. Rivière, G., Laîné, A., Lapeyre, G., Salas-Mélia, D., Kageyama, M., 2010. Links be- Hodell, D.A., Mead, G.A., Mueller, P.A., 1990. Variation in the strontium isotopic com- tween Rossby wave breaking and the North Atlantic Oscillation–Arctic Oscil- position of seawater (8 Ma to present): implications for chemical weathering lation in present-day and Last Glacial Maximum climate simulations. J. Cli- rates and dissolved fluxes to the oceans. Chem. Geol. 80, 291–307. mate 23, 2987–3008. Jouzel, J., Vaikmae, R., Petit, J.R., Martin, M., Duclos, Y., Stievenard, M., Lorius, C., Toots, M., Melieres, M.A., Burckle, L.H., Barkov, N.I., Kotlyakov, V.M., 1995. The Steig, E.J., Morse, D.L., Waddington, E.L., Stuiver, M., Grootes, P.M., Mayewski, P.A., two-step shape and timing of the last deglaciation in Antarctica. Clim. Dyn. 11, Twickler, M.S., Whitlow, S.I., 2000. Wisconsinan and Holocene climate history 151–161. from an ice core at Taylor Dome, western Ross Embayment, Antarctica. Geogr. Jweda, J., Bolge, L., Class, C., Goldstein, S.L., 2015. High Precision Sr–Nd–Hf–Pb Iso- Ann., Ser. A 82A (2–3), 213–235. topic Compositions of USGS Reference Material BCR-2. Geostand. Geoanal. Res. Stump, E., Fitzgerald, P.G., 1992. Episodic uplift of the Transantarctic Mountains. Ge- http://dx.doi.org/10.1111/j.1751-908X.2015.00342.x. ology 20 (2), 161–164. Kageyama, M., Valdes, P.J., 2000. Impact of the North American ice-sheet orography Uglietti, C., Gabrielli, P., Lutton, A., Olesik, J., Thompson, L., 2014. Large variability of on the Last Glacial Maximum eddies and snowfall. Geophys. Res. Lett. 27, 1515. trace element mass fractions determined by ICP-SFMS in ice core samples from Kawabe, I., Toriumi, T., Ohta, A., Miura, N., 1998. Monoisotopic REE abundances in worldwide high altitude glaciers. Appl. Geochem. 47, 109–121. seawater and the origin of seawater tetrad effect. Geochem. J. 32, 213–229. Vihma, T., 2014. Effects of Arctic Sea ice decline on weather and climate: a review. Koffman, B.G., Handley, M.J., Osterberg, E.C., Wells, M.L., Kreutz, K.J., 2014b. De- Surv. Geophys. 35 (5), 1175–1214. pendence of ice-core relative trace-element concentration on acidification. Wedepohl, K.H., 1995. The composition of the continental crust. Geochim. Cos- J. Glaciol. 60 (219), 103–112. mochim. Acta 59, 1217–1232. Koffman, B.G., Kreutz, K.J., Breton, D.J., Kane, E.J., Winski, D.A., Birkel, S.D., Kurbatov, Wegner, A., Gabrielli, P., Wilhelms-Dick, D., Ruth, U., Kriews, M., De Deckker, P., Bar- A.V., Handley, M.J., 2014a. Centennial-scale variability of the Southern Hemi- bante, C., Cozzi, G., Delmonte, B., Fischer, H., 2012. Change in dust variability in sphere westerly wind belt in the eastern Pacific over the past two millennia. the Atlantic sector of Antarctica at the end of the last deglaciation. Clim. Past 8 Clim. Past 10, 1125–1144. (1), 135–147. Koornneef, J.M., Bouman, C., Schwieters, J.B., Davies, G.R., 2014. Measurement of Wolff, E.W., Fischer, H., Fundel, F., Ruth, U., Twarloh, B., Littot, G.C., Mulvaney, R., small ion beams by thermal ionisation mass spectrometry using new 1013 Ohm Rothlisberger, R., de Angelis, M., Boutron, C.F., Hansson, M., Jonsell, U., Hutterli, resistors. Anal. Chim. Acta 819, 49–55. M.A., Lambert, F., Kaufmann, P., Stauffer, B., Stocker, T.F., Steffensen, J.P., Bigler, Laîné, A., Kageyama, M., Salas-Mélia, D., Voldoire, A., Rivière, G., Ramstein, G., Plan- M., Siggaard-Andersen, M.L., Udisti, R., Becagli, S., Castellano, E., Severi, M., Wa- ton, S., Tyteca, S., Peterschmitt, J.Y., 2008. Northern hemisphere storm tracks genbach, D., Barbante, C., Gabrielli, P., Gaspari, V., 2006. Southern Ocean sea-ice during the last glacial maximum in the PMIP2 ocean-atmosphere coupled mod- extent, productivity and iron flux over the past eight glacial cycles. Nature 440, els: energetic study, seasonal cycle, precipitation. Clim. Dyn. 32, 593–614. 491–496. Lambert, F., Bigler, M., Steffensen, J.P., Hutterli, M., Fischer, H., 2012. Centennial min- Wolff, E.W., Rankin, A.M., Röthlisberger, R., 2003. An ice core indicator of eral dust variability in high-resolution ice core data from Dome C, Antarctica. Antarctic sea ice production? Geophys. Res. Lett. 30. http://dx.doi.org/10.1029/ Clim. Past 8, 609–623. 2003gl018454.